Appropriate design of HELP is important particularly for novice users of software systems. Three methods of controlling HELP displays were evaluated in a text editing task. The three methods were the initiation of HELP (i.e., user vs. computer), The selection of a particular HELP sequence (i.e., user vs. computer), and the type of format (i.e., on-line vs. hard copy). These three variables were crossed in a between-subjects design along with a control condition of no HELP available to produce a total of nine separate test conditions. The various HELP configurations improved time to edit and reduced the quantity of errors and editing commands when compared to the no HELP control condition. Additionally, the efficacy of the various HELP configurations differed among themselves. It was concluded that for novice users, the optimum HELP system configuration was one in which the user controls the initiation and selection of HELP displays, and the information is stored in a hard-copy manual. The implications of this conclusion are discussed in terms of the design of HELP to facilitate browsing and comparison features.